Intel leaks show that next i7 may drop hyperthreading

Have a look at this review:

https://www.techspot.com/review/1613-amd-ryzen-2700x-2600x/

Handbrake 4K h.264 to 1080p h.265

Core i7-8700K @5.2GHz 16.4fps
Core i5-8600K @5.2GHz 13.7fps

Same # physical cores. Same CPU arch. Same clock speed. 6 threads vs 12 threads (6 threads + 6 hyperthreads)

Hyperthreaded beats non-hyperthreaded by 19.7%. Unless I'm reading your post wrong, this seems to be backwards of what you are claiming. And never mind about the Ryzen performance in this test.

I liked your post for sharing the link, but your statement is not 100% accurate. Do you look up the two chips' details? Those two intel cores are different. The 8700k has higher clock speed, significantly higher turbo speed, and more cache, which makes a significant difference. There are also other slight differences. If the only difference in price premium were hyperthreading, that would be a ripoff.

Take a look yourself:
https://www.intel.com/content/www/us/en/products/processors/core/core-vpro/i7-8700k.html
https://ark.intel.com/products/126685/Intel-Core-i5-8600K-Processor-9M-Cache-up-to-4_30-GHz

Both chips were overclocked to 5.2 GHz, not at their disparate native speeds. That (OK, in theory) should level the clock speed playing field as both chips are forced to run at the same speed, assuming no downclocking due to reaching their thermal limits. You're right about the cache differences between the 2 CPUs: 9MB vs. 12MB.

However, I'll bet that accounts for a lot less than the full 19.7% difference as I've done Handbrake encoding tests with older chips (quad core i7 Ivy Bridge), disabling and enabling hyperthreading and the speed improvement when encoding h.264 with hyperthreading on was about 11%. I still have that machine, I'll do a comparo at h.265 this evening and see if the difference has changed with newer HB versions and h.265.
 
As a tech and computer enthusiast I would hands down pick an i9 and pay the premium for it over any Ryzen CPU every single day of the week. Considering how close the 8700K gets to a 2700X in multithreaded performance I think it’s safe to assume the i9 is going to thrash the 2700X. I wouldn’t even be surprised if 8 cores without hyperthreading at 5ghz bears out the 2700X.

Until Ryzen 2 comes out its likely that Ryzen will be in the budget and value market. And I think that’s great for people who aren’t as sad as me who don’t care about having the best and want to make sure they can get a decent machine without breaking the bank. AMD have provided more options. Who knows with Ryzen 2 they might even make something I’d buy, if they can get an IPC boost strong enough to provide a strong single threaded upgrade to the current intel stuff then I’m all aboard. Im really not that bothered about “moar cores”. To me the core wars are beginning to remind me of the megapixel wars of digital cameras we had some years ago. It’s quality of the cores or pixels that count, not how many you have.
 
As a tech and computer enthusiast I would hands down pick an i9 and pay the premium for it over any Ryzen CPU every single day of the week. Considering how close the 8700K gets to a 2700X in multithreaded performance I think it’s safe to assume the i9 is going to thrash the 2700X. I wouldn’t even be surprised if 8 cores without hyperthreading at 5ghz bears out the 2700X.

2700X is 22% faster on Cinebench, so not very close. Intel CPU's are based on Pentium pro and so what? Meldown+Spectre+all upcoming point out why that matters. AMD also has some issues but situation is much better because architecture is much newer.

Until Ryzen 2 comes out its likely that Ryzen will be in the budget and value market. And I think that’s great for people who aren’t as sad as me who don’t care about having the best and want to make sure they can get a decent machine without breaking the bank. AMD have provided more options. Who knows with Ryzen 2 they might even make something I’d buy, if they can get an IPC boost strong enough to provide a strong single threaded upgrade to the current intel stuff then I’m all aboard. Im really not that bothered about “moar cores”. To me the core wars are beginning to remind me of the megapixel wars of digital cameras we had some years ago. It’s quality of the cores or pixels that count, not how many you have.

Ryzen's only real problem is manufacturing tech. GF 7nm tech should be much better than Intel's current so then AMD will dominate on everything finally.
 
2700X is 22% faster on Cinebench, so not very close. Intel CPU's are based on Pentium pro and so what? Meldown+Spectre+all upcoming point out why that matters. AMD also has some issues but situation is much better because architecture is much newer.



Ryzen's only real problem is manufacturing tech. GF 7nm tech should be much better than Intel's current so then AMD will dominate on everything finally.

Yeah people were saying Ryzen 1 will dominate, they were saying it for years, even when it was delayed and delayed. When it came out, it didn’t dominate, certainly it didn’t persuade gamers and overclockers and most of the market. Now we are waiting for Zen 2?
I was a spotty teenager the last time AMD dominated the market, it was a long long time ago. And they dominated with chips that were gaming champions but second best in productivity. You claim AMD are going to dominate next year? You claim that every year! Il believe it when I see it.
 
Yeah people were saying Ryzen 1 will dominate, they were saying it for years, even when it was delayed and delayed. When it came out, it didn’t dominate, certainly it didn’t persuade gamers and overclockers and most of the market. Now we are waiting for Zen 2?
I was a spotty teenager the last time AMD dominated the market, it was a long long time ago. And they dominated with chips that were gaming champions but second best in productivity. You claim AMD are going to dominate next year? You claim that every year! Il believe it when I see it.

Don't know where you got that info. Ryzen was expected to have Haswell level IPC and clocks around 3.2-3.4 GHz max because of 14nm LPP process. Somehow AMD got clocks to 4 GHz range, that's more than expected. For that reason Ryzen was never expected to be good for overclocking, never. Also Ryzen 1 was not delayed at all. It was launched 4.5 years after development began and that's Very fast. Normally it takes at least (that is, without delays) 5 years from development start to ready product. On paper GF 7nm tech is much better than Intel's 14nm tech so if specs hold true, Zen2 will easily dominate as Intel is struggling with 10nm tech.

You remember wrong. AMD dominated on everything on Athlon64 era.

I have never said that Ryzen was supposed to dominate so don't lie.
 
Don't know where you got that info. Ryzen was expected to have Haswell level IPC and clocks around 3.2-3.4 GHz max because of 14nm LPP process. Somehow AMD got clocks to 4 GHz range, that's more than expected. For that reason Ryzen was never expected to be good for overclocking, never. Also Ryzen 1 was not delayed at all. It was launched 4.5 years after development began and that's Very fast. Normally it takes at least (that is, without delays) 5 years from development start to ready product. On paper GF 7nm tech is much better than Intel's 14nm tech so if specs hold true, Zen2 will easily dominate as Intel is struggling with 10nm tech.

You remember wrong. AMD dominated on everything on Athlon64 I have never said that Ryzen was supposed to dominate so don't lie.

Read it and weep. Athlon 64 did not dominate everything;

https://www.anandtech.com/show/1164/16

“intel is still ahead in content creation applications”.

I remember right. You remember wrong. It’s those red tinted glasses you wear mate.
 
Last edited:
Congratulations of finding single Intel biased software. It's not hard to make software where AMD FX dominates every Intel CPU available.

That kind of software does not count for "everything".
Read all the tabs on it mate. It wasn’t just a single application....

“Intel continues to do extremely well under content creation applications such as DivX encoding; the clear leader here is still Intel”.

I’m strongly get the impression that you are rather ill informed of AMDs dominant era. It’s remarkable how AMD back then seem to be somewhat similar to how Intel are now. They even charged excruciatingly for their Athlon64-FX parts, with prices of over $1000, which back then was a lot more than it is now.
 
Read all the tabs on it mate. It wasn’t just a single application....

“Intel continues to do extremely well under content creation applications such as DivX encoding; the clear leader here is still Intel”.

I’m strongly get the impression that you are rather ill informed of AMDs dominant era. It’s remarkable how AMD back then seem to be somewhat similar to how Intel are now. They even charged excruciatingly for their Athlon64-FX parts, with prices of over $1000, which back then was a lot more than it is now.

AMD dominated on integer, FPU and especially memory performance. That can be considered everything. In fact, integrated memory controller alone can be considered overall domination as that affects on everything and makes use much more fluid. As I already stated, it's possible to make benchmark that gives better results for certain manufacturer's CPU. Like that http://arstechnica.com/reviews/hardware/atom-nano-review.ars/6

That shows AMD domination. Intel had to do that kind of tricks because AMD was miles better in every aspect: price, performance and heat.

For DivX software, there are some "oddities" https://www.extremetech.com/computing/55036-athlon-64-amd-plays-its-trump-card/9

Clearly something was badly wrong with 3200+ so I wouldn't consider DivX tests to have any value.

Intel priced hotter and slower Extreme Edition $999. So why should AMD price faster and cooler CPU any cheaper?
 
Yeah, I'm thinking about who suffers from this most. I would say "creative arts" and "finance" gain the most from hyper-threading. Certainly not going to affect gaming so why hamper those kinds of users? Seems like a very "off target" decision?
 
Have a look at this review:

https://www.techspot.com/review/1613-amd-ryzen-2700x-2600x/

Handbrake 4K h.264 to 1080p h.265

Core i7-8700K @5.2GHz 16.4fps
Core i5-8600K @5.2GHz 13.7fps

Same # physical cores. Same CPU arch. Same clock speed. 6 threads vs 12 threads (6 threads + 6 hyperthreads)

Hyperthreaded beats non-hyperthreaded by 19.7%. Unless I'm reading your post wrong, this seems to be backwards of what you are claiming. And never mind about the Ryzen performance in this test.

I liked your post for sharing the link, but your statement is not 100% accurate. Do you look up the two chips' details? Those two intel cores are different. The 8700k has higher clock speed, significantly higher turbo speed, and more cache, which makes a significant difference. There are also other slight differences. If the only difference in price premium were hyperthreading, that would be a ripoff.

Take a look yourself:
https://www.intel.com/content/www/us/en/products/processors/core/core-vpro/i7-8700k.html
https://ark.intel.com/products/126685/Intel-Core-i5-8600K-Processor-9M-Cache-up-to-4_30-GHz

Both chips were overclocked to 5.2 GHz, not at their disparate native speeds. That (OK, in theory) should level the clock speed playing field as both chips are forced to run at the same speed, assuming no downclocking due to reaching their thermal limits. You're right about the cache differences between the 2 CPUs: 9MB vs. 12MB.

However, I'll bet that accounts for a lot less than the full 19.7% difference as I've done Handbrake encoding tests with older chips (quad core i7 Ivy Bridge), disabling and enabling hyperthreading and the speed improvement when encoding h.264 with hyperthreading on was about 11%. I still have that machine, I'll do a comparo at h.265 this evening and see if the difference has changed with newer HB versions and h.265.

OK, followup:

Core i7-3720QM 4 cores @2.6GHz, turbo boost off to avoid any thermal throttling. ~10 minute 1080p/30 h.264 video Handbrake convert to h.265 rf20, medium, mp3 passthru.

4 cores, no hyperthreads: 9.21 fps
4 cores, plus 4 hyperthreads: 10.57 fps

That's a 14.8% increase in speed with hyperthreading on. This is consistent with my previous experiences using hyperthreading with Handbrake - it gives higher performance as long as your cooling solution can handle it.

Relevant to that:

4 cores, no hyperthreads: 19.8W average
4 cores, plus 4 hyperthreads: 22.1W average

4 cores, no hyperthreads: 82C average
4 cores, plus 4 hyperthreads: 84C average

5500rpm fan (max), 26C ambient temp. This is a 2012 Mac Mini FWIW.
 
AMD dominated on integer, FPU and especially memory performance. That can be considered everything. In fact, integrated memory controller alone can be considered overall domination as that affects on everything and makes use much more fluid. As I already stated, it's possible to make benchmark that gives better results for certain manufacturer's CPU. Like that http://arstechnica.com/reviews/hardware/atom-nano-review.ars/6

That shows AMD domination. Intel had to do that kind of tricks because AMD was miles better in every aspect: price, performance and heat.

For DivX software, there are some "oddities" https://www.extremetech.com/computing/55036-athlon-64-amd-plays-its-trump-card/9

Clearly something was badly wrong with 3200+ so I wouldn't consider DivX tests to have any value.

Intel priced hotter and slower Extreme Edition $999. So why should AMD price faster and cooler CPU any cheaper?
I’m not denying AMD dominated mate, they absolutely did. I’m calling out when they say they dominated everything. You are absolutely wrong on that. Content creators and business users we’re better on Intel at the time. And I didn’t claim that Intel didn’t also charge high prices, they did. Stop straw-manning.

Nothing badly wrong with the 3200+, it was a great CPU, better than most Intel CPUs at the time for gamers. It just doesn’t match your rather amusing delusion that AMD were the best at everything.
 
I’m not denying AMD dominated mate, they absolutely did. I’m calling out when they say they dominated everything. You are absolutely wrong on that. Content creators and business users we’re better on Intel at the time. And I didn’t claim that Intel didn’t also charge high prices, they did. Stop straw-manning.

Nothing badly wrong with the 3200+, it was a great CPU, better than most Intel CPUs at the time for gamers. It just doesn’t match your rather amusing delusion that AMD were the best at everything.

Business software? AMD was much better there and because of integrated memory controller, difference is indeed much bigger than benchmarks say. You are saying Intel was "better" at content creation because of single DivX software. As I provided link, you can see there is much wrong with DivX (3200+ result is flawed) so that can be safely ignored. Not to mention Intel's power consumption was so huge that AMD could just raise clocks a little and left Intel more behind. AMD didn't need to do that.

Also for content creation generally, it made use of multiple CPU's at that time. So those who did content creation used multiple CPU configurations. On dual CPU (or quad CPU) configurations AMD was even more ahead because of non-shared bus of Hyper Transport.

So basically AMD was best on everything except few Intel optimized software and some cases that were rarely used (like media encoding on single CPU).
 
OK, followup:

Core i7-3720QM 4 cores @2.6GHz, turbo boost off to avoid any thermal throttling. ~10 minute 1080p/30 h.264 video Handbrake convert to h.265 rf20, medium, mp3 passthru.

4 cores, no hyperthreads: 9.21 fps
4 cores, plus 4 hyperthreads: 10.57 fps

That's a 14.8% increase in speed with hyperthreading on. This is consistent with my previous experiences using hyperthreading with Handbrake - it gives higher performance as long as your cooling solution can handle it.

Relevant to that:

4 cores, no hyperthreads: 19.8W average
4 cores, plus 4 hyperthreads: 22.1W average

4 cores, no hyperthreads: 82C average
4 cores, plus 4 hyperthreads: 84C average

5500rpm fan (max), 26C ambient temp. This is a 2012 Mac Mini FWIW.

Thank you so much for doing that! This is confusing why Intel would not feature HT on new CPUs? Unless there are some tests they have run where it is detrimental? Maybe they are holding it back to release the next wave of CPUs with HT as "faster"?
 
Business software? AMD was much better there and because of integrated memory controller, difference is indeed much bigger than benchmarks say. You are saying Intel was "better" at content creation because of single DivX software. As I provided link, you can see there is much wrong with DivX (3200+ result is flawed) so that can be safely ignored. Not to mention Intel's power consumption was so huge that AMD could just raise clocks a little and left Intel more behind. AMD didn't need to do that.

Also for content creation generally, it made use of multiple CPU's at that time. So those who did content creation used multiple CPU configurations. On dual CPU (or quad CPU) configurations AMD was even more ahead because of non-shared bus of Hyper Transport.

So basically AMD was best on everything except few Intel optimized software and some cases that were rarely used (like media encoding on single CPU).

You really need to read the review mate, or any review of the Athlon64. It wasn’t just divX that Intel were stronger at, a lot of other applications were better on Intel at the time. The Athlon64 did dominate because they were faster overall and faster for gaming and yes I’m sure had better power efficiency (although I would check, you are incredibly biased towards AMD and not trustworthy at all). You are factually incorrect if you believe that AMD dominated everything, this is easy to find out. You clearly weren’t into this stuff during that time as you are clearly quite mis informed on the subject. Of course I hugely preferred AMD back then as I’m a gamer and I prefer chips that cost a bit more but perform better. This was the Athlon64, it was expensive (so were Intel). The AthlonXP before it were the good value parts. I owned both a Pentium 4 and an Athlon64 and I massively preferred the Athlon64 for my needs. I miss that AMD, even if the prices were sky high.

I will point out, if Zen 2 comes out and offers a significant IPC advantage over Intel and can clock as high as Intel chips I fully believe they will charge more than Intel are charging, it’s how this industry works. Mostly because people like me will be prepared to pay it. Or best case scenario they will cost the same as Intel and Intel will cut their prices and AMD won’t. I wouldn’t mind that arrangement, we haven’t seen a decent improvement in IPC for years, I’d love to see one, I don’t care if it’s from AMD or Intel, whoever does it gets my money at this point.

Read through the benchmarks on this review, there are quite a few that Pentium were faster at; https://techreport.com/review/5683/amd-athlon-64-processor/13

To quote the article;

“For those of us with more pedestrian spending limits, the Athlon 64 3200+ looks like a great value. Yes, it costs over 400 bucks, but the stock Pentium 4 3.2GHz is selling for more than $600 right now. The Athlon 64 3200+ maybe trails the P4 3.2GHz in overall performance by the thinnest of margins, but no way is the P4 worth another $150 to $200. And that's without considering the 64-bit question.”.

When AMD released its more powerful Athlon 64 parts they ranked up their prices and Intel cut theirs leading to the value crown going to Intel at least for non gamers. Quote from an Athlon64 4000 review;

“In fact, once you take price into account, the Pentium 4s start looking quite a bit more attractive than they might otherwise. Throw out the P4 Extreme Edition; the Prescott-based P4 560 is often faster than the Extreme Edition, and the 560 lists for only $417. The 560 is also much cheaper than the Athlon 64 3800+, currently priced at $643. AMD may be beating Intel in performance, but the value proposition isn't there until you work your way down to the 3500+. Even among mid-range processors, outside of gaming performance, the Pentium 4 and Athlon 64 offerings are generally competitive, as the WorldBench scores testify. The P4 550 3.4GHz outscores the Athlon 64 3500+, and the P4 540 at 3.2GHz edges out the Athlon 64 3200+. So Intel may be down, but it's not completely out of contention, especially for non-gamers.”.

I had the Athlon64 3500+ which replaced my P4 3.0 from the previous Prescott generation and was quite significantly better at gaming but not so at productivity. It also cost more.
 
Last edited:
You really need to read the review mate, or any review of the Athlon64. It wasn’t just divX that Intel were stronger at, a lot of other applications were better on Intel at the time. The Athlon64 did dominate because they were faster overall and faster for gaming and yes I’m sure had better power efficiency (although I would check, you are incredibly biased towards AMD and not trustworthy at all). You are factually incorrect if you believe that AMD dominated everything, this is easy to find out. You clearly weren’t into this stuff during that time as you are clearly quite mis informed on the subject. Of course I hugely preferred AMD back then as I’m a gamer and I prefer chips that cost a bit more but perform better. This was the Athlon64, it was expensive (so were Intel). The AthlonXP before it were the good value parts. I owned both a Pentium 4 and an Athlon64 and I massively preferred the Athlon64 for my needs. I miss that AMD, even if the prices were sky high.

Read through the benchmarks on this review, there are quite a few that Pentium were faster at; https://techreport.com/review/5683/amd-athlon-64-processor/13

To quote the article;

“For those of us with more pedestrian spending limits, the Athlon 64 3200+ looks like a great value. Yes, it costs over 400 bucks, but the stock Pentium 4 3.2GHz is selling for more than $600 right now. The Athlon 64 3200+ maybe trails the P4 3.2GHz in overall performance by the thinnest of margins, but no way is the P4 worth another $150 to $200. And that's without considering the 64-bit question.”.

In that article they used whopping 14 applications. Of those 14 applications, at least 3ds max, Lightwave, LAME, DivX and Quake 3 arena are well known to be Intel biased applications. Quake 3 is not exactly biased. It runs so well that ID didn't bother to fix broken 3D-NOW! support because it was more than fast enough for everybody. However that shows in benchmarks. They also didn't test processor thermals where AMD wins hands down. Finally, those benchmarks contain software that were not at all optimized for Athlon64, whereas there were many optimizations for Pentium 4.

For benchmarks, there is cache issue on Pentium 4. What happens when CPU has not enough cache to fit everything? Then CPU uses memory. Pentium 4 memory accesses were via chipset but AMD had integrated memory controller on CPU. What happens when there are multiple software on same time that all consume cache or cache pollution happens? Intel suffers a lot, AMD suffers much less. That's why those benchmarks don't really tell whole picture.

There is also clock speed issue. When manufacturer dominates on CPU field, that dominating manufacturer simply checks what competitor have and puts out CPU that is fast enough to beat competition but not too fast because that affects profits. Those Athlon64's were "only" 2.2 GHz parts whereas Intel parts were 3.2 GHz. Intel never released faster Pentium 4 than 3.8 GHz (+0.6 GHz) while AMD released 3.2 GHz Athlon64 (+1 GHz). Also AMD never had heat problems with Athlon64 but there were serious problems with Intel Prescott's. AMD just didn't need to put out any faster Athlon64 because those were more than fast enough for everything except some rare cases AMD didn't consider worthy.

Today Intel quad cores (like i7-4770K) are considered "better at everything" vs AMD Bulldozer quad cores like FX-8350. However there are many software out where FX-8350 beats i7-4770K. Few software won't change big picture and it's practically impossible to build CPU that is faster than competitor on everything if that everything means 100% of all software.

To summarize: At that time AMD could easily launch CPU that trounce Intel even on those Intel biased benchmarks. That would have been bad for profits. Just like couple of years ago. AMD stopped high end FX-series development to FX-8350 so Intel decided to release at most low clock quad cores into desktop. There were still some software where FX was faster but Intel didn't care about those because i7 was faster on virtually everything and profits were higher that way. In both cases it's safe to say dominating CPU was faster at everything.

I will point out, if Zen 2 comes out and offers a significant IPC advantage over Intel and can clock as high as Intel chips I fully believe they will charge more than Intel are charging, it’s how this industry works. Mostly because people like me will be prepared to pay it. Or best case scenario they will cost the same as Intel and Intel will cut their prices and AMD won’t. I wouldn’t mind that arrangement, we haven’t seen a decent improvement in IPC for years, I’d love to see one, I don’t care if it’s from AMD or Intel, whoever does it gets my money at this point.

No. AMD can never charge more than Intel for similar part because Intel brand is so much stronger. Weaker brand must offer same quality for lower price or better quality for same price. Otherwise better brand always wins. This has been very evident when AMD has been competing against Intel. Brand value is main reason why many times newcomers have much difficulty breaking into market even if newcomers have better product in every aspect.

IPC development has been pretty slow lately. Intel's latest development was Skylake, that makes it three years without any development. After Ryzen AMD has also made zero improvement. Ryzen 2 had some IPC improvement thanks to faster caches, but those faster caches existed on Ryzen "1". For some reason (yields, manufacturing etc) AMD slowed Ryzen "1" caches and Ryzen 2 has cache speeds they first intended. There is hope that Zen 2 brings considerable IPC improvement. Ryzen was rushed to market and borrowed some parts from FX Excavator. Zen 2 should contain only totally new parts, not any borrowed ones.
 
Thank you so much for doing that! This is confusing why Intel would not feature HT on new CPUs? Unless there are some tests they have run where it is detrimental? Maybe they are holding it back to release the next wave of CPUs with HT as "faster"?

I'm been receiving errors trying to post a longer response but in short, I think Intel is saving hyperthreading on high core count CPUs for the top end parts. Which is apparently the i9 nowadays.
 
I'm been receiving errors trying to post a longer response but in short, I think Intel is saving hyperthreading on high core count CPUs for the top end parts. Which is apparently the i9 nowadays.
To be honest, the only way this makes sense (if the rumours are true), is from an economics point of view. Intel sure as hell doesn't want to cannibalize their expensive high-end desktop/i9 lineup, but since they probably don't have anything better to offer at this point, they'd rather sacrifice hyperthreading in their i7 processors for now... and milk their high IPC to death.
 
Back